Shape property cooperative control method for magnesium alloy plate single-pass controlled rolling forming

Through the finite element coupled VPSC physical model and machine learning data-driven model, the problems of low forming efficiency and poor stamping forming in magnesium alloy sheet rolling technology are solved, and the rapid and precise rolling control of magnesium alloy sheets are achieved, and the material performance is improved and utilization.

CN119016499AActive Publication Date: 2024-11-26JIMEI UNIV

Patent Information

Application Number
CN202410949942.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-26
Estimated Expiration
2044-07-16

AI Technical Summary

Technical Problem

The magnesium alloy sheet rolling technology has problems such as low forming efficiency, high anisotropy, poor stamping forming ability and serious edge cracking, which limits its application in transportation, aerospace and other fields.

Method used

Using finite element coupled VPSC physical model and machine learning data-driven model, the relationship model of rolling process parameters-organization parameters-static/dynamic mechanical properties is constructed through FEM-VPSC simulation and machine learning model to realize single-pass control of magnesium alloy plates Fast and precise control of the rolling forming process.

Benefits of technology

Real-time and rapid interaction of the rolling process of magnesium alloy sheets is realized, the rolling process parameters and deformation process are optimized and controlled, the mechanical properties of the structure are regulated, the forming efficiency and material utilization are improved, and the risk of edge cracking is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119016499A_ABST
    Figure CN119016499A_ABST
Patent Text Reader

Abstract

The invention provides a magnesium alloy plate single-pass controlled rolling forming shape property cooperative control method which is characterized by comprising the following operations: S1, establishing an FEM-VPSC physical model of a finite element coupling viscoplastic self-consistent model; s2, establishing a machine learning data driving model; s3, performing shape cooperative control based on the constructed FEM-VPSC physical model and the machine learning data driving model; based on an FEM-VPSC physical model, a physical relation model among rolling process parameters, structure parameters and static / dynamic mechanical property parameters is built, and forward prediction and control are carried out; a data driving model among rolling process parameters, structure parameters and static / dynamic mechanical property parameters is built based on a data driving model of machine learning, forward prediction and control are carried out, reverse prediction and control are carried out, array parameters, rolling process parameters and a deformation process are optimized and controlled, the structure mechanical property is regulated and controlled, and the rolling quality is improved. And rapid and accurate control of shape cooperation is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for coordinated shape control of single-pass controlled rolling forming of a magnesium alloy plate. Background Art

[0002] Magnesium alloy is the lightest metal structural material and has broad application prospects in the fields of transportation, aerospace, communications and electronics. However, magnesium alloys have bottlenecks such as poor strength-ductility matching and low forming efficiency, which limit their practical application. Rolling is one of the main methods for producing magnesium alloy sheets. It can flexibly produce sheets of different thicknesses and widths with high forming efficiency. The rolling process changes the rolling process and controls the evolution of the organization to obtain the target mechanical properties.

[0003] Rolling can refine grains, improve structure and significantly improve the mechanical properties of alloys. However, the current magnesium alloy sheet rolling technology is still immature. The rolled sheets have high anisotropy, poor stamping formability, severe edge cracking, and low material utilization. Therefore, it is necessary to promote the development of magnesium alloys by studying the rolling method of magnesium alloy sheets. Summary of the invention

[0004] The present invention proposes a method for coordinated shape control of single-pass controlled rolling forming of a magnesium alloy plate, the purpose of which is to achieve rapid and accurate control of the single-pass controlled rolling forming process of the magnesium alloy plate, which is specifically achieved by the following technical means: A method for coordinated shape control of single-pass controlled rolling of a magnesium alloy plate, comprising the following operations: S1, Establishment of FEM-VPSC simulation model of finite element coupled viscoplastic self-consistent model The FEM-VPSC simulation model is used to simulate the single-pass controlled rolling forming process and the static / dynamic service deformation of the controlled rolled magnesium alloy plate; The simulation of the above-mentioned single-pass controlled rolling forming process includes: 1) Taking material parameters, rolling temperature and rolling average strain rate as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field and unit velocity gradient during array controlled rolling forming; 2) The unit velocity gradients at different geometric positions of the formed workpiece calculated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macro deformation law and texture evolution in the forming process, and a relationship model between rolling process parameters and organizational parameters is established. The process parameters include rolling temperature and rolling average strain rate, and the organizational parameters include texture pole density. The above-mentioned simulation of static / dynamic service deformation of the controlled rolled magnesium alloy plate includes: 1) Taking material parameters, organizational parameters, and static / dynamic load parameters as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field, and unit velocity gradient during static / dynamic service deformation; 2) The unit velocity gradients at different geometric positions of the workpiece during the static / dynamic service deformation process simulated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macroscopic deformation law and texture evolution in the static / dynamic service deformation process, and the static / dynamic mechanical performance index parameters in the static / dynamic service deformation process are predicted, and a physical relationship model of the organizational parameters-static / dynamic mechanical performance parameters is established, wherein the organizational parameters include the texture extreme density, and the static / dynamic mechanical performance parameters include the peak stress; S2, Building Machine Learning Data-Driven Models The machine learning method is used to construct a data-driven model of rolling process parameters, organizational parameters, and static / dynamic mechanical properties parameters, including: 1) removing outliers from the rolling process parameters, organizational parameters, and static / dynamic mechanical property parameters related to the controlled-rolled magnesium alloy plate, and then normalizing the data; the rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress; 2) Use multiple inputs / outputs to build a machine learning model. After selection, the data set is randomly divided into a training set and a prediction set according to M:N, where M and N are preset natural numbers and M is greater than or equal to N; the above-mentioned multiple inputs / outputs include: A. Select rolling process parameters as input parameters and organizational parameters as output parameters; B. Select rolling process parameters as input parameters and mechanical property parameters as output parameters; C. Select rolling process parameters and organizational parameters as input parameters, and mechanical property parameters as output parameters; 3) Construct a machine learning model, determine the algorithm parameters in the machine learning model, train the machine learning model with training samples, and predict the organizational parameters and performance parameters of the controlled rolled magnesium alloy plate after training; 4) Analyze the prediction results of the machine learning model, select the machine learning model with the best prediction performance among the machine learning algorithms used, and use it to reversely determine the microstructure parameters and rolling process parameter ranges that can produce controlled-rolled magnesium alloy plates with better performance parameters; S3, shape-coordinated control based on the constructed FEM-VPSC physical model and machine learning data-driven model; Based on the FEM-VPSC physical model, a physical relationship model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters is constructed for forward prediction and control; A data-driven model based on machine learning is used to build a data-driven model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters to perform forward prediction and control, as well as reverse prediction and control; The rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress.

[0005] Compared with the prior art, the advantages of the present invention are as follows: VPSC physical model, machine learning data-driven model, and data streams such as rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters in the model can interact quickly and in real time during the single-pass rolling forming process and static / dynamic service process of magnesium alloy plates. Through the forward prediction and control of the VPSC physical model and the forward / reverse prediction and control of the machine learning data-driven model, the rolling process parameters and deformation process can be optimized and controlled, the organizational mechanical properties can be regulated, and the rapid and precise control of the synergy of the form and properties can be achieved to form a technical prototype. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 It is a collaborative control flow chart based on extended VPSC multi-scale simulation and machine learning.

[0007] Figure 2 The numerical maps of texture pole density and peak stress corresponding to data of different process parameters. DETAILED DESCRIPTION

[0008] Combine the following Figure 1 and 2 , further describe this application scheme: A method for coordinated shape control of single-pass controlled rolling of a magnesium alloy plate, comprising the following operations: S1, establish the FEM-VPSC simulation model of the finite element (FEM) coupled viscoplastic self-consistent model (VPSC) The FEM-VPSC simulation model is used to simulate the single-pass controlled rolling forming process and the static / dynamic service deformation of the controlled rolled magnesium alloy plate; The simulation of the above-mentioned single-pass controlled rolling forming process includes: 1) Taking material parameters, rolling temperature and rolling average strain rate as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field and unit velocity gradient during array controlled rolling forming; 2) The unit velocity gradients at different geometric positions of the formed workpiece calculated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macro deformation law and texture evolution in the forming process, and a relationship model between rolling process parameters and organizational parameters is established. The process parameters include rolling temperature and rolling average strain rate, and the organizational parameters include texture pole density. The above-mentioned simulation of static / dynamic service deformation of single-pass controlled rolling magnesium alloy plate includes: 1) Taking material parameters, organizational parameters, and static / dynamic load parameters as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field, and unit velocity gradient during static / dynamic service deformation; 2) The unit velocity gradients at different geometric positions of the workpiece during the static / dynamic service deformation process simulated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macroscopic deformation law and texture evolution in the static / dynamic service deformation process, and the static / dynamic mechanical performance index parameters in the static / dynamic service deformation process are predicted, and a physical relationship model of the organizational parameters-static / dynamic mechanical performance parameters is established, wherein the organizational parameters include the texture extreme density, and the static / dynamic mechanical performance parameters include the peak stress; S2, Building Machine Learning Data-Driven Models The data-driven model of rolling process parameters, organizational parameters and static / dynamic mechanical properties parameters is constructed using machine learning methods including artificial neural network (ANN), random forest (RF), support vector machine (SVR), K-nearest neighbor (KNN) and extreme gradient boosting (XGboost), including: 1) removing outliers from the rolling process parameters, organizational parameters, and static / dynamic mechanical property parameters of the single-pass controlled rolling magnesium alloy plate, and then normalizing the data; the rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress; 2) Use multiple inputs / outputs to build a machine learning model. After selection, the data set is randomly divided into a training set and a prediction set according to M:N. Here, M and N are preset natural numbers and M is greater than or equal to N. For example, M is 7 and N is 3, that is, after selection, the data set is randomly divided into a training set and a prediction set according to 7:3. The above multiple inputs / outputs include: A. Select rolling process parameters as input parameters and organizational parameters as output parameters; B. Select rolling process parameters as input parameters and mechanical property parameters as output parameters; C. Select rolling process parameters and organizational parameters as input parameters, and mechanical property parameters as output parameters; 3) Construct a machine learning model, determine the algorithm parameters in the machine learning model, train the machine learning model with training samples, and predict the organizational parameters and performance parameters of the controlled rolled magnesium alloy plate after training; 4) Analyze the prediction results of the machine learning model, select the machine learning model with the best prediction performance among the adopted machine learning algorithms, and use it to reversely determine the microstructure parameters and rolling process parameter ranges of the single-pass controlled-rolled magnesium alloy plate with better performance parameters; S3, shape-coordinated control based on the constructed FEM-VPSC physical model and machine learning data-driven model; Based on the FEM-VPSC physical model, a physical relationship model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters is constructed for forward prediction and control; A data-driven model based on machine learning is used to build a data-driven model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters to perform forward prediction and control, as well as reverse prediction and control; The rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress.

[0009] The present invention establishes a finite element coupled VPSC physical model and a machine learning data driven model, and the model finally obtains a numerical map of texture pole density and peak stress corresponding to data of different process parameters (see Appendix Figure 2 ), in the graph, curve A represents the average rolling strain rate that changes with temperature, curve B represents the texture extreme density that changes with temperature, and curve C represents the peak stress that changes with temperature; based on the relationship presented in the graph, the data streams such as rolling process parameters, organizational parameters, static / dynamic mechanical properties parameters in the model can interact in real time and quickly during the single-pass rolling forming process and static / dynamic service process of the magnesium alloy plate. Through the forward prediction and control of the VPSC physical model and the forward / reverse prediction and control of the machine learning data-driven model, the rolling process parameters and deformation process can be optimized and controlled, the organizational mechanical properties can be regulated, and the rapid and precise control of the synergy of form and properties can be achieved.

[0010] The above-mentioned preferred implementation modes should be regarded as illustrative examples of the implementation modes of the present application scheme. Any technical deductions, replacements, improvements, etc. that are identical or similar to the present application scheme or made based on this scheme should be regarded as within the scope of protection of this patent.

Claims

1. A method for coordinated shape control of single-pass controlled rolling of magnesium alloy plates, characterized in that: The following operations are included: S1, Establishment of FEM-VPSC simulation model of finite element coupled viscoplastic self-consistent model The FEM-VPSC simulation model is used to simulate the single-pass controlled rolling forming process of magnesium alloy plates and the static / dynamic service deformation of single-pass controlled rolled magnesium alloy plates; The simulation of the above-mentioned single-pass controlled rolling forming process includes: 1) Taking material parameters, rolling temperature and rolling average strain rate as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field and unit velocity gradient during single-pass controlled rolling forming; 2) The unit velocity gradients at different geometric positions of the formed workpiece calculated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macro deformation law and texture evolution in the forming process, and a relationship model between rolling process parameters and organizational parameters is established. The process parameters include rolling temperature and rolling average strain rate, and the organizational parameters include texture pole density. The above-mentioned simulation of static / dynamic service deformation of the controlled rolled magnesium alloy plate includes: 1) Taking material parameters, organizational parameters, and static / dynamic load parameters as input, the FEM simulation model is used to simulate the temperature field, stress field, strain field, and unit velocity gradient during static / dynamic service deformation; 2) The unit velocity gradients at different geometric positions of the workpiece during the static / dynamic service deformation process simulated by the FEM model are loaded into the VPSC model as output, and the twin recrystallization and grain boundary bowing recrystallization micromechanics models are implanted into the VPSC model; 3) The VPSC model is used to simulate the macroscopic deformation law and texture evolution in the static / dynamic service deformation process, and the static / dynamic mechanical performance index parameters in the static / dynamic service deformation process are predicted, and a physical relationship model of the organizational parameters-static / dynamic mechanical performance parameters is established, wherein the organizational parameters include the texture extreme density, and the static / dynamic mechanical performance parameters include the peak stress; S2, Building Machine Learning Data-Driven Models The machine learning method is used to construct a data-driven model of rolling process parameters, organizational parameters, and static / dynamic mechanical properties parameters, including: 1) removing outliers from the rolling process parameters, organizational parameters, and static / dynamic mechanical property parameters of the single-pass controlled rolling magnesium alloy plate, and then normalizing the data; the rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress; 2) Use multiple inputs / outputs to build a machine learning model. After selection, the data set is randomly divided into a training set and a prediction set according to M:N, where M and N are preset natural numbers and M is greater than or equal to N; the above-mentioned multiple inputs / outputs include: A. Select rolling process parameters as input parameters and organizational parameters as output parameters; B. Select rolling process parameters as input parameters and mechanical property parameters as output parameters; C. Select rolling process parameters and organizational parameters as input parameters, and mechanical property parameters as output parameters; 3) Construct a machine learning model, determine the algorithm parameters in the machine learning model, train the machine learning model with training samples, and predict the organizational parameters and performance parameters of the controlled rolled magnesium alloy plate after training; 4) Analyze the prediction results of the machine learning model, select the machine learning model with the best prediction performance among the adopted machine learning algorithms, and use it to reversely determine the microstructure parameters and rolling process parameter ranges of the single-pass controlled-rolled magnesium alloy plate with better performance parameters; S3, shape-coordinated control based on the constructed FEM-VPSC physical model and machine learning data-driven model; Based on the FEM-VPSC physical model, a physical relationship model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters is constructed for forward prediction and control; A data-driven model based on machine learning is used to build a data-driven model between rolling process parameters, organizational parameters, and static / dynamic mechanical performance parameters to perform forward prediction and control, as well as reverse prediction and control; The rolling process parameters include rolling temperature and rolling average strain rate; the organizational parameters include texture pole density; and the static / dynamic mechanical property parameters include peak stress.

2. The method for coordinated shape control of single-pass controlled rolling of magnesium alloy sheet according to claim 1, characterized in that: The machine learning method includes one or more of artificial neural network, random forest, support vector machine, K-nearest neighbor, and extreme gradient boosting.

3. The method for coordinated shape control of single-pass controlled rolling of magnesium alloy sheet according to claim 1, characterized in that: In step S2, the natural number M is taken as 7, and the natural number N is taken as 3, that is, the selected data set is randomly divided into a training set and a prediction set in a ratio of 7:3.

Citation Information

Patent Citations

  • Large-strain extrusion cutting strip texture prediction method considering strain rate influence

    CN115422797A

  • Mechanism collaborative forecasting method and system for forecasting mechanical properties of whole-process hot-rolled strip steel

    CN115815345A

  • Additive manufacturing part stress prediction method based on crystal plasticity finite element method

    CN118298974A

  • VPSC model construction method based on ABAQUS sub-model

    CN120656610A

  • Display device and method of driving the same

    KR102750835B1

Cited By

  • VPSC model construction method based on ABAQUS sub-model

    CN120656610A

  • Fatigue life simulation evaluation method for lightweight aluminum alloy material of new energy automobile

    CN120995753A